Research On Model And Algorithm Of Optimal Order Qutatities For Determining Demand

With increasing global competition, customer-centric supply chain management model is gradually replaced the traditional management model with production and product-centric,the prodect of "Manufacturing-distribution\wholesale-retail" process and the process of participating physical activities and relationship coordination are the main task of supply chain management.Purchasing management is one of the most important components of the supply chain management,playing an increasing important role in value creation and growth to the whoel supply chain. The enterprise efficient purchasing management can not only reduce their cost and enhance market competitivenss, but also create their own new competitiveness,so as to ensure the fast chaning market are in a leading position.First, review the theory of theinventory management and the procurementmanagement. Analyses and summarizes one enterprise’s characteristics,proposed the problem"How to allocate the order quantities in the multiple order cycle at the same time order a varieties of goods".The problem considers the determining demand based on the downstream customer orders,the minimum order quantity limited,Storage capacity and liquidity constraints, to minimize the operating costs.Established two optimization models which one is based on the purchase amount discount under determining demand to determine the order quantity and the otheroptimization model is based on purchase quantity discounts in the same condition.Secondly, design and implementation of genetic algorithm and particle swarm optimization algorithmfor solving the proposed model.The genetic algorithm design includes the heuristic method to initialize the population, two crossover methods which one is uniform crossover using gene segment and the orther is sample crossover under gene segment, two mutation methods which one is simple mutation under gene segment and the orther is heuristic mutation under gene segment and penalty strategy and repairing strategy to process the infeasible chromosomes. The particle swarm optimization algorithm design includeseach dimension of each particle is designed to be the order of different kinds of goods in each cycle,the heuristic method to initialize the swarm, dynamic adjustment of learning factor during particle update speed and position, random weight, ring structure based on an index number with a shortcut and penalty strategy to process the infeasible chromosomes. Finally, the algorithms are implemented with C#high-level language programming.Application case is given to perform the simulation study and sensitivity analysis, and to verify the feasibility of the proposed model and algorithms. The algorithm experiments includes the parameter simulation of the genetic algorithm and particle swarm optimization, the performance analysis of different crossover and mutation operators of genetic algorithm, the comparison of penalty strategy and repairing strategy and comparative analysis of the performance of genetic algorithm and particle swarm optimization algorithm.The model simulation experiment includes parameter sensitivity analysis of the inventory capacity, liquidity, different types of discounts, as well as the key constraint on the order quantitiy. Simulation results validate the effectiveness and feasibility of the model and algorithms.